Last edit:  Sept. 2003

GOAL: Combine decision analysis with NPV to evaluate a problem of choice under uncertainty.


You still work for a major materials manufacturer.  Your boss examined the preliminary NPV analysis,
but has some concerns.  While initial orders of two million parts/year are guaranteed, there is some
uncertainty regarding the future rate of demand growth.  Your initial assessment used a single growth forecast.  
You have now been asked to consider the implications if the growth rate is twice the estimate,
or if no growth occurs.  Further details are provided below.


Same basic details as in Optimal Plant Investment, Part 1.
Initial demand always 2 million parts/ year.
Demand growth rate uncertain; for each of the following years it can follow one of these scenarios:
    High: 3 million parts/year;
    Medium: 1.5 million parts/year;
    Low: 0 parts/year.
Each growth scenario has an equal likelihood of occurring (probability = 1/3).
Maximum demand remains 10 million parts/ year.
Use a discount rate of 10 percent and consider 10 years of sales.


(1) You should refer to the spreadsheet you developed for Part 1.   For each possible growth rate,
calculate the NPV for Plan A, Plan B with a capacity addition in Year 3, and Plan B without a capacity
addition in Year 3.
Calculate the expected NPV for each investment plan (probability weighted average).

(2) Build a decision tree which compares Plan A and Plan B (with and without the capacity expansion
decided at Year 0).  Here the expansion decision is assumed to be fixed in advance.

(3) Revise the decision tree to reflect the fact that the expansion decision is made in Year 3.
Build a decision tree that can be used to estimate the value of a test which predicts demand growth with
certainty for this case of expansion being decided in Year 3.  In short, calculate the expected value of
perfect information, as described in the class presentations.
Your report should include the three decision trees and a brief (one page) summary with
recommendation and discussion of issues.

Additionally, note that the overall expected demand in any given year is exactly that estimated in Part 1
(i.e. expected growth rate equals 1.5 million parts per year).   However, the expected NPV calculated in
Part 2
does not equal the NPV calculated using the expected growth rate.  Explain this observation,
and its implications (briefly).



(see TreeAge Manual or Instructions Summary for more detail)

When you create active links between TreeAge and your spreadsheet, make sure that you have gathered
all your results on the same sheet and that you have named the cells.
If you enter the probabilities as variables, you will be able to perform sensitivity analyses on these uncertainties.
TreeAge enables you to check for your EVPI calculation.